TWI902572B - Wafer and roughness improvement method, device, equipment and medium - Google Patents

Abstract

This application provides a method, apparatus, device, and medium for improving the roughness of a wafer; the method includes: detecting the roughness of a bare wafer after a wire dicing process in a target wavelength range; when the roughness is greater than or equal to a set roughness threshold, selecting dicing process parameters that need to be adjusted to reduce the roughness of the bare wafer obtained in subsequent wire dicing processes in the target wavelength range based on the priority of dicing process parameters indicated by a fitting function between multiple dicing process parameters and the roughness in the target wavelength range.

Description

Methods, apparatus, equipment and dielectrics for improving wafers and their roughness

This application claims priority to Chinese Patent Application No. 202411435849.1, filed in China on October 15, 2024, the entire contents of which are incorporated herein by reference.

This application relates to the field of semiconductor manufacturing technology, and more particularly to a method, apparatus, equipment and dielectric for improving the roughness of a wafer.

In the wafer manufacturing process, after single-crystal silicon ingots are prepared using the Czochralski method, they undergo a series of processing steps including wire cutting, grinding, etching, polishing, chemical mechanical polishing (CMP), and cleaning to ultimately obtain single-crystal silicon wafers. For single-crystal silicon wafers, surface morphology is a key parameter for evaluating their quality. Surface roughness is an important performance parameter among wafer surface morphology parameters.

Based on the various processing steps in the wafer manufacturing process described above, after the single-crystal silicon ingot is cut into a bare wafer using a wire dicing machine, the bare wafer already possesses an initial surface condition. Subsequent processing steps such as grinding, etching, polishing, CMP, and cleaning merely serve to gradually repair surface damage based on this initial surface condition. Therefore, if a single-crystal silicon wafer with better surface roughness is desired, process optimization needs to be initiated during the wire dicing process to improve the surface roughness of the bare wafer obtained from the wire dicing process, thereby enhancing the final surface roughness index of the single-crystal silicon wafer.

However, there is currently no solution to optimize the wire dicing process based on the surface roughness of the bare wafer obtained from the wire dicing process.

This application provides a method, apparatus, equipment, and dielectric for improving the surface roughness of a wafer; using the roughness of the bare wafer obtained by wire cutting within a target wavelength range as an evaluation index, and adjusting the process parameters in the wire cutting process through this index to reduce the surface roughness of the wafer.

The technical solution of this application is implemented as follows:

In a first aspect, this application provides a method for improving wafer roughness, the method comprising:

Detect the roughness of a bare wafer after wire cutting in the target wavelength range;

When the roughness is greater than or equal to a set roughness threshold, the dicing process parameters that need to be adjusted are selected based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the roughness in the target wavelength range, so as to reduce the roughness of the bare wafer obtained by the subsequent wire dicing process in the target wavelength range.

Secondly, this application provides a wafer roughness improvement device, the device comprising: a detection section, a comparison section, and a process improvement section; wherein,

The detection section is configured to detect the roughness of the bare wafer after the wire cutting process in the target wavelength range;

The comparison section is configured to compare the roughness with a set roughness threshold, and to trigger the process improvement section when the roughness is greater than or equal to the set roughness threshold.

The process improvement section is configured to select the dicing process parameters that need to be adjusted to reduce the roughness of the bare wafer obtained from subsequent wire dicing processes in the target wavelength range, based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the roughness in the target wavelength range.

Thirdly, this application provides a computing device comprising: a processor and a memory; the processor being configured to execute instructions stored in the memory to implement the wafer roughness improvement method as described in the first aspect.

Fourthly, this application provides a computer-readable storage medium storing at least one instruction, which is executed by a processor to implement the wafer roughness improvement method as described in the first aspect.

Fifthly, this application provides a wafer obtained by a wire dicing process improved by the wafer roughness improvement method described in the first aspect, and the roughness in the target wavelength range is at least one of the following:

When the target wavelength range is 0 to 1.8 micrometers, the roughness is 0.01 to 1 nm;

When the target wavelength range is 1.8 to 22 micrometers, the roughness is 1 to 3 nm;

When the target wavelength range is 22 micrometers to 20 millimeters, the roughness is 3 to 30 nm.

This application provides a method, apparatus, equipment, and dielectric for improving the roughness of a wafer and the wafer thereof. Using the roughness of a bare wafer obtained by wire dicing within a target wavelength range as an evaluation index, when the roughness of the bare wafer after wire dicing is greater than or equal to a set roughness threshold within the target wavelength range, the method selects the dicing process parameters that need adjustment based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the roughness within the target wavelength range. The method reduces the roughness of the bare wafer obtained in subsequent wire dicing processes within the target wavelength range by adjusting the selected dicing process parameters.

The technical solution of this application will be clearly and completely described below with reference to the accompanying drawings.

Referring to Figure 1, an exemplary multi-wire cutting apparatus 1 applicable to the technical solutions of this application is illustrated. The multi-wire cutting apparatus 1 may include a wire cutting unit 11 and a support unit 12. In some embodiments of this application, the wire cutting unit 11 may be located below the support unit 12 in the vertical direction. In embodiments not shown in this application, the wire cutting unit 11 may also be located above the support unit 12 in the vertical direction, which will not be elaborated here.

The wire EDM unit 11 may include multiple spools 111, cutting wires 112, and a cutting fluid supply unit 113. The cutting wires 112 are wound around the spools 111 to form an array of parallel cutting segments. The following description uses two spools 111 as an example, with reference to FIG1. The two spools 111 are arranged opposite each other with their axes parallel to each other. Two cutting fluid supply units 113 are located directly above the two spools 111, respectively, and are used to spray cutting fluid onto the two spools 111 and the cutting wires 112. The cutting fluid may be a liquid medium in which abrasive particles are suspended.

The reciprocating motion directions of the spool 111 and the cutting wire 112 toward and away from the load-bearing unit 12 are shown by the solid arrows in Figure 1. The reciprocating motion speed can, for example, be from 5 m/s to 18 m/s. The load-bearing unit 12 is used to load and fix the silicon rod S to be processed, and may include a base 121 and an intermediate member 122. The silicon rod S to be processed is fixed to the base 121 via the intermediate member 122, for example, the silicon rod to be processed can be fixed to the base 121 by bonding its circumferential surface to the lower surface of the base 121.

For the multi-wire cutting equipment 1 shown in Figure 1, the silicon rod to be processed can be brought close to the cutting segment array by moving the support unit 12 vertically toward the cutting segment array. After the cutting segment array comes into contact with the silicon rod to be processed, the silicon rod S is cut by the movement of each cutting segment array along its extension direction and the continued feed movement of the support unit 12 vertically toward the cutting segment array. It should be noted that the movement of the wire support unit 12 is achieved by a lifting device (not shown in the figure). It is understood that those skilled in the art can also achieve the movement of the support unit 12 in other ways according to actual needs and implementation scenarios, which will not be described in detail in this application.

Multiple grooves 111A are formed on the circumferential surface of the spindle 111 to guide and fix the cutting wire 112, ensuring that it maintains stable tension and position during the cutting process. Typically, the grooves 111A are evenly distributed on the circumferential surface of the spindle 111.

During the wire cutting process based on the aforementioned multi-wire cutting equipment 1, the cutting fluid cannot be sprayed onto the middle part of the silicon rod S to be processed. The cutting fluid for cutting this part can only be obtained by being carried and transported by the cutting wire. Since the cutting wire can only operate in a limited space, and there is no strong adhesive relationship between the cutting fluid and the cutting wire material, the cutting fluid attached to the cutting wire is easily scraped off during the transportation of the cutting wire. As a result, when cutting the middle part of the silicon rod S to be processed, the content of cutting fluid is less than that of the edge part of the silicon rod S to be processed. Insufficient cutting fluid content will lead to a decrease in the cutting force along the cutting direction. Specifically, due to the lack of a convection heat dissipation path for the cutting fluid, more heat will accumulate in the middle part of the silicon rod S to be processed and cannot be dissipated. This will cause thermal stress differences throughout the entire wafer area, resulting in wafer deformation.

Furthermore, as the silicon rod S is fed downwards, the cutting area during the wire EDM process undergoes a process of increasing and decreasing. Taking the cross-section of the silicon rod shown in Figure 2 as an example, as the silicon rod S is fed downwards as indicated by the arrow, the cutting area gradually increases from a chord (dashed line) to a diameter and then decreases back to a chord (dotted line). Based on Figure 2, when the cutting area is a diameter, more heat accumulates inside the bare wafer, resulting in a higher risk of thermal stress deformation.

Based on the above, the inventors discovered that maintaining a constant cutting rate may not be an ideal wire EDM solution. As the cutting area increases, the process parameters related to the cutting rate can be appropriately adjusted to enable more adequate heat dissipation inside the silicon rod S to be processed, thereby improving the roughness of the bare wafer.

Based on this, referring to Figure 3, which illustrates a method for improving wafer roughness provided in this application, the method comprising steps S301 to S302.

In step S301, the roughness of the bare wafer after the wire dicing process is detected in the target wavelength range.

In this application, the uniformity of the wafer surface morphology is reflected in the height difference between points on the wafer surface. This height difference manifests as a fluctuation in the height value of each point on the wafer surface as a whole. Based on this understanding, this application considers this fluctuation phenomenon from a wave perspective as being caused by the superposition of wave signals of different wavelengths. The smaller the wavelength of the wave signal, the smaller the area of height value fluctuation phenomenon it represents; the larger the wavelength of the wave signal, the larger the area of height value fluctuation phenomenon it represents.

Based on the above explanation, in order to evaluate or measure the roughness of even smaller regions, this application can filter the aforementioned fluctuation phenomenon using the size of the region where the roughness is to be measured, thereby obtaining wave signal data within the wavelength range corresponding to the region where the roughness is to be measured, and calculating the wafer roughness under that region size standard based on the wave signal data. In this application, a target wavelength range of 22 micrometers to 20 millimeters is used as an example.

In step S302, when the roughness is greater than or equal to a set roughness threshold, the dicing process parameters that need to be adjusted are selected based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the roughness in the target wavelength range, so as to reduce the roughness of the bare wafer obtained by the subsequent wire dicing process in the target wavelength range.

In this application, after detecting the roughness of the bare wafer after the wire dicing process within the target wavelength range, the roughness can be characterized by a set roughness threshold. For example, with a target wavelength range of 22 micrometers to 20 millimeters, in some examples, this roughness threshold can be set to 25 nm. That is, when the roughness value of the bare wafer within the target wavelength range is greater than or equal to 25 nm, it indicates that the cutting capability of the multi-wire dicing equipment 1 is severely degraded.

Specifically, the bare wafers obtained from the wire dicing process exhibit obvious periodic lines on a macroscopic scale. These lines can be considered as a general representation of roughness on the wafer surface. Based on this understanding, it can be known that these lines, like the roughness mentioned in the previous steps, can exhibit a wave phenomenon with amplitude, and this wave phenomenon is also caused by the superposition of wave signals of different wavelengths.

Based on the structure and working principle of the multi-wire EDM equipment shown in Figure 1, the inventors discovered that the macroscopic appearance (i.e., the wire marks) is related to the following cutting process parameters used in the wire EDM process: the moving speed of the cutting wire reciprocating motion, the cycle of the cutting wire reciprocating motion, and the feed speed of the silicon rod S to be processed. The degree of influence of the above three cutting process parameters on the surface roughness of the bare wafer in the target wavelength range can be characterized by a fitting function. In this application, the fitting function is represented by a vis function of the above three cutting process parameters, for example, as shown in the following formula:

in, This indicates the speed at which the cutting line moves back and forth. Vf represents the period of the reciprocating motion of the cutting line, and Vf represents the feed rate. b and c are empirical coefficients in the fitting function, representing the degree of influence of their respective dicing process parameters on the surface roughness. K1 represents the fitting coefficient. To determine the empirical and fitting coefficients of the fitting function, in some examples, dicing process parameters collected from historical wire EDM processes and the surface roughness measured on bare wafers obtained from historical wire EDM processes within the target wavelength range are substituted into the fitting function to determine the empirical coefficients in the fitting function. b , c, and K1 .

In this application, K1 = 6.2 × ^10⁻³ . = 0.529, b = 1.445, c = 0.873, based on this empirical coefficient The specific values of b and c show that the period of the reciprocating motion of the cutting line corresponding to the empirical coefficient b has the greatest impact on the surface roughness, followed by the feed rate corresponding to the empirical coefficient c , and then the empirical coefficient itself. The corresponding reciprocating motion speed of the cutting wire. The magnitude of the above influence is correspondingly prioritized for the reciprocating motion speed, the cycle of the cutting wire reciprocating motion, and the feed rate, respectively. Specifically, the greater the influence, the higher the priority. Therefore, when the roughness exceeds the set roughness threshold, the cutting process parameters are adjusted according to the above priorities, thereby reducing the roughness of the bare wafer obtained in subsequent wire cutting processes within the target wavelength range.

The technical solution shown in Figure 3 above uses the roughness of the bare wafer obtained by wire cutting within the target wavelength range as an evaluation index. When the roughness of the bare wafer after wire cutting is greater than or equal to the set roughness threshold within the target wavelength range, the dicing process parameter that needs to be adjusted is selected based on the priority of the dicing process parameter indicated by the fitting function between multiple dicing process parameters and the roughness within the target wavelength range. The roughness of the bare wafer obtained by subsequent wire cutting processes within the target wavelength range is reduced by adjusting the selected dicing process parameter.

For the technical solution shown in Figure 3, in some possible implementations, the fitting function can be obtained by fitting the cutting process parameters collected in the historical wire cutting process and the roughness in the target wavelength range detected on the bare wafer obtained from the historical wire cutting process. The specific fitting process may include:

An initial fitting function is formed by multiplying the sine function of the reciprocating motion of the cutting line with the first empirical coefficient as the exponent, the sine function of the reciprocating motion of the cutting line with the second empirical coefficient as the exponent, and the sine function of the feed rate with the third empirical coefficient as the exponent.

In the historical wire EDM process, historical EDM process parameters are collected, and the roughness of the bare wafer that has completed the historical wire EDM process is detected in the target wavelength range.

The historical dicing process parameters and the roughness of the bare wafer in the target wavelength range after completing the historical wire dicing process are substituted into the fitting function to obtain the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function.

For the above implementation method, exemplary specific implementation details may include:

First, a fitting function is derived from the power functions of the three cutting process parameters mentioned above, as shown below:

Where Vs represents the moving speed of the cutting line reciprocating motion, Tr represents the period of the cutting line reciprocating motion, and Vf represents the feed rate. b and c are empirical coefficients in the fitting function, which respectively represent the degree of influence of their corresponding cutting process parameters on the roughness. K1 represents the fitting coefficient.

Next, the above fitting function is logarithmically transformed to obtain the logarithmically transformed fitting function as shown in the following equation:

.

Then, using the historical dicing process parameters collected in the historical wire dicing process and the roughness of the bare wafer in the target wavelength range after the historical wire dicing process is detected, the empirical coefficient and the fitting coefficient are calculated by the least squares method for the fitting function after the logarithmic transformation, thereby obtaining the fitting function that can be applied to the technical solution shown in Figure 3.

Through the above process of obtaining the fitting function, we can finally obtain K1 = 6.2 × ^10⁻³ . = 0.529, b = 1.445, c = 0.873. Based on the above empirical coefficients, it can be concluded that the period of the reciprocating motion of the cutting line has the greatest impact on the surface roughness, followed by the feed rate, and then the moving speed of the reciprocating motion of the cutting line.

In this application, some historical dicing process parameters and the corresponding roughness of bare wafers that have completed historical wire dicing processes in the range of 22 micrometer wavelength to 20 millimeter wavelength are shown in Table 1 below. Vf (mm/min) Vs (m/s) Tr (s) Roughness in the target wavelength range 3 15 80 38.1 3 15 65 28.2 3 15 40 14 2 15 40 9.8 1.5 15 40 7.6 1.5 12 40 6.8 1.5 8 40 5.5

Table 1

In this application, when the target wavelength range is 1.8 to 22 micrometers, based on the above fitting function acquisition process, K1 = 4.4 × ^10⁻³ can be obtained. = 0.659, b = 1.217, c = 0.887. Based on the above empirical coefficients, it can also be concluded that the period of the reciprocating motion of the dicing wire has the greatest impact on the roughness, followed by the feed rate, and then the moving speed of the reciprocating motion of the dicing wire. In this application, some historical dicing process parameters and the corresponding roughness of bare wafers that have completed historical wire dicing processes in the wavelength range of 1.8 to 22 micrometers are shown in Table 2 below.

Vf (mm/min) Vs (m/s) Tr (s) Roughness in the target wavelength range 3 15 40 6.2 3 15 30 4.4 3 15 20 2.7 2 15 20 1.9 1.5 15 20 1.4 1.5 12 20 1.2 1.5 8 20 1.0

Table 2

In this application, when the target wavelength range is 0 to 1.8 micrometers, based on the above fitting function acquisition process, K1 = 6.7 × ^10⁻³ can be obtained. = 0.813, b = 1.105, c = 0.971. Based on the above empirical coefficients, it can also be concluded that the period of the reciprocating motion of the dicing wire has the greatest impact on the roughness, followed by the feed rate, and then the moving speed of the reciprocating motion of the dicing wire. In this application, some historical dicing process parameters and the corresponding roughness of bare wafers that have completed historical wire dicing processes in the wavelength range of 0 to 1.8 micrometers are shown in Table 3 below. Vf (mm/min) Vs (m/s) Tr (s) Roughness in the target wavelength range 1.5 8 20 1.5 1.5 8 15 1.1 1.5 8 10 0.7 1 8 10 0.5 0.7 8 10 0.3 0.7 6 10 0.3 0.7 4 10 0.2

Table 3

Tables 1, 2, and 3 above also verify the conclusions regarding the influence of the cycle of the cutting line reciprocating motion, the feed rate, and the moving speed of the cutting line reciprocating motion on the roughness, based on the changes in roughness with variations in Vf, Vs, and Tr.

For the technical solution shown in Figure 3, in some possible implementations, as shown in Figure 4, the process of detecting the roughness of the bare wafer in the target wavelength range after the wire cutting process described in step S301 may include steps S3011 to S3013.

In step S3011, for each sampling point on the surface of the bare wafer, raw measurement data about the wafer surface height at each sampling point is obtained using a single-point measurement scheme.

Specifically, in this implementation, a single-point measurement scheme means that the measurement data can only be obtained at one sampling point in a single measurement process. In some examples, a contact measurement scheme can be used, such as using a probe to contact the surface of a bare wafer and move horizontally across the wafer surface. With this horizontal movement, the height difference on the wafer surface causes a longitudinal displacement of the probe. This longitudinal displacement is sensed by a displacement sensor, and the sensed longitudinal displacement value is converted into the height data of the wafer surface, i.e., the raw measurement data about the wafer surface height. In other examples, non-contact measurement schemes such as capacitance measurement or laser focusing measurement can also be used.

In step S3012, based on the raw measurement data of all sampling points on the surface of the bare wafer, the wave signal of each sampling point in the target wavelength range is obtained for each sampling point through a filter corresponding to the target wavelength range.

In this implementation, after measuring each sampling point on the bare wafer surface using the aforementioned single-point measurement scheme, the raw measurement data of all sampling points can reflect the height difference between each sampling point, which can be viewed as a three-dimensional wave phenomenon on the entire wafer surface. For this wave phenomenon, the amplitude of the wave signals of different wavelengths superimposed to form the wave phenomenon can be used to represent the height difference within the region corresponding to the wavelength. Taking a target wavelength range of 22 micrometers to 20 millimeters as an example, the upper limit of the wavelength is 20 mm, and the lower limit is 22 μm. Taking a target wavelength range of 0 to 1.8 micrometers as an example, the upper limit of the wavelength is 1.8 μm, and the lower limit is 0. Taking a target wavelength range of 1.8 to 22 micrometers as an example, the upper limit of the wavelength is 22 μm and the lower limit of the wavelength is 1.8 μm.

In some examples, in order to obtain the wave signal of each sampling point within the target wavelength range, in the specific implementation of step S3012, firstly, a low-pass filter function is used to filter out wave signals with wavelengths higher than the upper limit of the wavelength from the wave phenomenon, and then a high-pass filter function is used to filter out wave signals with wavelengths lower than the lower limit of the wavelength, finally obtaining the wave signal within the target wavelength range.

Specifically, taking the aforementioned filter as an example, which is a circular filter composed of two Gaussian filtering functions, the first Gaussian filtering function _GLP1 can be a low-pass Gaussian filtering function whose low-pass filtering range covers to the upper limit of the target wavelength range; the second Gaussian filtering function _GLP2 can be a low-pass Gaussian filtering function whose low-pass filtering range covers to the lower limit of the target wavelength range, and the high-pass filtering function is obtained through (1- _GLP2 ). Based on the aforementioned first and second Gaussian filtering functions, the circular filter _GDHP can be expressed as _GDHP = _GLP1 (1- _GLP2 ). In this application, taking the circular filter of the above example as an example, as shown in Figure 5, the circular filter indicated by the arrow can be slid on the surface of the bare wafer according to the sampling points. The original measurement data of the sampling point at the center of the circular filter and other sampling points covered by the effective range of the circular filter are filtered by the circular filter to obtain the wave signal of each sampling point in the target wavelength range.

It should be noted that the morphology changes more drastically closer to the edge of the wafer surface. To accurately capture these dramatic morphological changes, the effective range of a circular filter corresponding to a sampling point near the wafer edge should be smaller than that of a circular filter corresponding to a sampling point closer to the wafer center. In other words, as the sampling point moves further away from the wafer center, the effective range of the corresponding filter should decrease, or shrink. Therefore, the radius of the filter's effective range is determined by the distance between the sampling point and the center of the bare wafer.

In this application, a critical distance is set to determine whether the sampling point is close to the center of the wafer surface or close to the edge of the wafer surface. When the distance between the sampling point and the center of the bare wafer is less than or equal to the critical distance, the radius of the effective range of the filter is a first radius; when the distance between the sampling point and the center of the bare wafer is greater than the critical distance, the radius of the effective range of the filter is a second radius; wherein, both the first radius and the second radius are determined by the upper limit of the target wavelength range, and the first radius is greater than the second radius.

Based on the above example, the shrinkage can be implemented in a stepwise manner, where the second radius is fixed at a value smaller than the first radius when the distance between the sampling point and the center of the bare wafer is greater than the critical distance; or it can be implemented gradually, where the second radius gradually decreases as the sampling point moves further away from the center of the bare wafer. Specifically, when the distance between the sampling point and the center of the bare wafer is greater than the critical distance, the second radius is a fixed value smaller than the first radius; or, the second radius is negatively correlated with the distance between the sampling point and the center of the bare wafer.

In step S3013, the roughness of the bare wafer in the target wavelength range is determined based on the amplitude statistics of the wave signals at all sampling points in the target wavelength range.

For the above implementation, after obtaining the wave signals of all sampling points within the target wavelength range through filtering, in some examples, determining the roughness of the bare wafer within the target wavelength range based on the amplitude statistics of the wave signals of all sampling points within the target wavelength range includes:

The bare wafer is divided into multiple analysis regions;

The full range of the amplitude of the wave signal in the target wavelength range of each analysis area is calculated based on the amplitude value of the wave signal at the sampling point in each analysis area.

The average amplitude of the wave signal across all analysis regions within the target wavelength range is determined as the roughness index of the bare wafer within the target wavelength range.

For the example above, as shown in Figure 6, after the edge region of the bare wafer is removed according to EE, the remaining region is divided into square regions of a set size, such as 10mm×10mm. This results in a complete analysis region that can present a complete square and a non-complete analysis region that is at the edge of the remaining region after the edge region is removed according to EE and cannot present a complete square. These two analysis regions constitute the analysis region in the above implementation method.

It should be noted that the wave signals of all sampling points on the wafer surface within the target wavelength range include all information that can characterize the surface morphology (i.e., roughness) of the wafer surface within the corresponding size region of the target wavelength range. This information can be characterized by statistical values of amplitude, such as the average value, extreme values, extreme values under a set size area ratio, range, variance, standard deviation, median, mode, etc., of the amplitude mentioned in the aforementioned technical solutions.

In this application, taking the full range as an example, the amplitude statistics of the wave signal in the target wavelength range for each analysis region are the full range of the amplitude values of the wave signal in the target wavelength range for the sampling points and interpolation points in each analysis region; correspondingly, the roughness index of the bare wafer in the target wavelength range is the average value of the full range of the amplitude values of the wave signal in the target wavelength range for all analysis regions.

Based on the above technical solutions and specific embodiments, the wafer obtained by the wire dicing process after improvement by the aforementioned wafer roughness improvement method has a roughness in the target wavelength range of at least one of the following:

When the target wavelength range is 0 to 1.8 micrometers, the roughness is 0.01 to 1 nm;

When the target wavelength range is 1.8 to 22 micrometers, the roughness is 1 to 3 nm;

When the target wavelength range is 22 micrometers to 20 millimeters, the roughness is 3 to 30 nm.

Based on the same inventive concept as the aforementioned technical solution, referring to Figure 7, a wafer roughness improvement device 70 is shown. The improvement device 70 includes: a detection section 701, a comparison section 702, and a process improvement section 703; wherein,

The detection section 701 is configured to detect the roughness of the bare wafer after the wire cutting process in the target wavelength range.

The comparison section 702 is configured to compare the roughness with a set roughness threshold, and to trigger the process improvement section 703 when the roughness is greater than or equal to the set roughness threshold.

The process improvement section 703 is configured to select the dicing process parameters that need to be adjusted to reduce the roughness of the bare wafer obtained in the target wavelength range in subsequent wire dicing processes, based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and roughness in the target wavelength range.

In some examples, the plurality of cutting process parameters include: the moving speed of the cutting wire reciprocating motion, the period of the cutting wire reciprocating motion, and the feed rate;

Correspondingly, the fitting function is represented by the product of sine functions relating to the moving speed of the cutting line reciprocating motion, the period of the cutting line reciprocating motion, and the feed rate.

In some examples, referring to FIG8, the improvement device 70 further includes: a fitting portion 704, configured to:

An initial fitting function is formed by multiplying the sine function of the reciprocating motion of the cutting line with the first empirical coefficient as the exponent, the sine function of the reciprocating motion of the cutting line with the second empirical coefficient as the exponent, and the sine function of the feed rate with the third empirical coefficient as the exponent.

In the historical wire EDM process, historical EDM process parameters are collected, and the roughness of the bare wafer that has completed the historical wire EDM process is detected in the target wavelength range.

The historical dicing process parameters and the roughness of the bare wafer in the target wavelength range after completing the historical wire dicing process are substituted into the fitting function to obtain the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function.

In some examples, the proposed coupling portion 704 is configured to:

The fitting function is logarithmically transformed to obtain the logarithmically transformed fitting function;

Using the historical dicing process parameters and the roughness of the bare wafer after the historical wire dicing process in the target wavelength range, the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function after logarithmic transformation are calculated by the least squares method.

In some examples, the detection section 701 is configured to:

For each sampling point on the surface of the bare wafer, raw measurement data about the wafer surface height at each sampling point is obtained using a single-point measurement scheme;

Based on the raw measurement data of all sampling points on the surface of the bare wafer, the wave signal of each sampling point in the target wavelength range is obtained by a filter corresponding to the target wavelength range for each sampling point;

The roughness of the bare wafer in the target wavelength range is determined based on the amplitude statistics of the wave signals at all sampling points in the target wavelength range.

In some examples, the filter is a circular filter and consists of a double Gaussian filtering function;

The radius of the filter's effective range is determined by the distance between the sampling point and the center of the bare wafer. When the distance between the sampling point and the center of the bare wafer is less than or equal to the critical distance, the radius of the filter's effective range is a first radius; when the distance between the sampling point and the center of the bare wafer is greater than the critical distance, the radius of the filter's effective range is a second radius. Both the first and second radii are determined by the upper limit of the target wavelength range, and the first radius is greater than the second radius.

In some examples, the detection section 701 is configured to:

The bare wafer is divided into multiple analysis regions;

The full range of the amplitude of the wave signal in the target wavelength range of each analysis area is calculated based on the amplitude value of the wave signal at the sampling point in each analysis area.

The average amplitude of the wave signal across all analysis regions within the target wavelength range is determined as the roughness index of the bare wafer within the target wavelength range.

In some examples, the roughness threshold is 1 nm when the target wavelength range is 0 to 1.8 micrometers.

When the target wavelength range is 1.8 to 22 micrometers, the roughness threshold is 3 nm;

When the target wavelength range is 22 micrometers to 20 millimeters, the roughness threshold is 25 nm.

It should be noted that for the specific implementation of the functions configured in each "part" of the above device, please refer to the implementation method and examples of the corresponding steps in the aforementioned wafer roughness improvement method, which will not be repeated here.

Please refer to Figure 9, which shows a structural block diagram of a computing device provided in an exemplary embodiment of this application. In some examples, the computing device 90 may be at least one of a smartphone, smartwatch, desktop computer, laptop computer, virtual reality terminal, augmented reality terminal, wireless terminal, and laptop computer. The computing device 90 has communication capabilities and can access a wired or wireless network. The computing device 90 may refer to one of a plurality of terminals, and those skilled in the art will understand that the number of such terminals may be more or less. In some examples, the computing device 90 may receive data based on the accessed wired or wireless network. Understandably, the computing device 90 undertakes the calculation and processing work of the technical solution of this application, and this application does not limit it in this regard.

As shown in Figure 9, the computing device in this application may include one or more of the following components: processor 910 and memory 920.

Optionally, the processor 910 connects to various parts of the computing device using various interfaces and wiring, and performs various functions of the computing device and processes data by running or executing instructions, programs, code sets or instruction sets stored in memory 920, and by calling data stored in memory 920. Optionally, the processor 910 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). Processor 910 can integrate one or more of the following: a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Neural-network Processing Unit (NPU), and a baseband chip. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content displayed on the touchscreen; the NPU implements artificial intelligence (AI) functions; and the baseband chip handles wireless communication. It is understood that the baseband chip can also be implemented as a separate chip without being integrated into processor 910.

The memory 920 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 920 may include a non-transitory computer-readable storage medium. The memory 920 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 920 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above method embodiments, etc.; the data storage area may store data created according to the use of the computing device, etc.

In addition, those skilled in the art will understand that the structure of the computing device shown in the above figures does not constitute a limitation on the computing device. The computing device may include more or fewer components than shown, or combine certain components, or arrange different components. For example, the computing device may also include a screen, camera module, microphone, speaker, RF circuit, input unit, sensors (such as accelerometer, angular velocity sensor, light sensor, etc.), audio circuit, WiFi module, power supply, Bluetooth module, etc., which will not be described in detail here.

This application also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor to implement the wafer roughness improvement methods described in the above embodiments.

This application also provides a computer program product including computer instructions stored in a computer-readable storage medium; a processor of a computing device reads the computer instructions from the computer-readable storage medium, executes the computer instructions, and causes the computing device to perform methods for improving wafer roughness as described in the above embodiments.

This application also provides a wafer obtained by a wire dicing process improved by the wafer roughness improvement method described in the foregoing technical solution, and having a roughness in the target wavelength range of at least one of the following:

When the target wavelength range is 0 to 1.8 micrometers, the roughness is 0.01 to 1 nm;

When the target wavelength range is 1.8 to 22 micrometers, the roughness is 1 to 3 nm;

When the target wavelength range is 22 micrometers to 20 millimeters, the roughness is 3 to 30 nm.

Those skilled in the art will appreciate that, in one or more of the examples above, the functions described in this application can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one location to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

It should be noted that the technical solutions described in this application can be combined arbitrarily without conflict.

The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the aforementioned claims.

1: Multi-wire EDM equipment 11: Wire EDM unit 111: Wire spool 111A: Wire groove 112: Cutting wire 113: Cutting fluid supply unit 12: Support unit 121: Base 122: Intermediate component 70: Improvement device 701: Detection section 702: Comparison section 703: Process improvement section 704: Fitting section 90: Calculation equipment 910: Processor 920: Memory S: Silicon rod to be processed S301: Detect the roughness of the bare wafer after the wire EDM process in the target wavelength range S3011: For each sampling point on the surface of the bare wafer, obtain the raw measurement data of the wafer surface height at each sampling point through a single-point measurement scheme S3012: Based on the raw measurement data of all sampling points on the surface of the bare wafer, the wave signal of each sampling point in the target wavelength range is obtained through a filter corresponding to the target wavelength range for each sampling point S3013: Based on the amplitude statistics of the wave signals of all sampling points in the target wavelength range, the roughness of the bare wafer in the target wavelength range is determined S302: When the roughness is greater than or equal to a set roughness threshold, the dicing process parameters that need to be adjusted are selected according to the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the roughness in the target wavelength range, so as to reduce the roughness of the bare wafer obtained by the subsequent wire dicing process in the target wavelength range.

Figure 1 is a schematic diagram of an exemplary multi-wire cutting device provided in this application.

Figure 2 is a schematic diagram of the cutting area during the wire EDM process provided in this application.

Figure 3 is a schematic flowchart of a method for improving wafer roughness provided in this application.

Figure 4 is a flowchart illustrating the detection of roughness in the target wavelength range provided in this application.

Figure 5 is a schematic diagram of a filter provided in this application sliding on the wafer surface.

Figure 6 is a schematic diagram of an analysis region obtained by dividing the region according to the present application.

Figure 7 is a schematic diagram of a wafer roughness improvement device provided in this application.

Figure 8 is a schematic diagram of another wafer roughness improvement device provided in this application.

Figure 9 is a schematic diagram of the structure of a computing device provided in this application.

S301: Detect the surface roughness of the bare wafer after the wire dicing process in the target wavelength range. S302: When the surface roughness is greater than or equal to a set surface roughness threshold, select the dicing process parameters that need adjustment to reduce the surface roughness of the bare wafer obtained in subsequent wire dicing processes, based on the priority of the dicing process parameters indicated by the fitting function between multiple dicing process parameters and the surface roughness in the target wavelength range.

Claims (12)

A method for improving wafer roughness, the method comprising: detecting the roughness of a bare wafer after a wire dicing process in a target wavelength range; when the roughness is greater than or equal to a set roughness threshold, selecting dicing process parameters that need adjustment based on the priority of dicing process parameters indicated by a fitting function between multiple dicing process parameters and the roughness in the target wavelength range. As described in claim 1, wherein the plurality of cutting process parameters include: the moving speed of the cutting wire reciprocating motion, the period of the cutting wire reciprocating motion, and the feed rate; correspondingly, the fitting function is represented by the product of sigma functions relating to the moving speed of the cutting wire reciprocating motion, the period of the cutting wire reciprocating motion, and the feed rate, respectively. The method of claim 1, further comprising: Forming an initial fitting function based on the product of a sigma function with respect to the moving speed of the wire reciprocating motion, exponentiated by a first empirical coefficient; a sigma function with respect to the period of the wire reciprocating motion, exponentiated by a second empirical coefficient; and a sigma function with respect to the feed rate, exponentiated by a third empirical coefficient; In a historical wire dicing process, acquiring historical dicing process parameters and detecting the roughness of the bare wafer after the historical wire dicing process in the target wavelength range; Substituting the historical dicing process parameters and the roughness of the bare wafer after the historical wire dicing process in the target wavelength range into the fitting function to obtain the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function. The method described in claim 3, wherein substituting the historical dicing process parameters and the roughness of the bare wafer after the historical wire dicing process in the target wavelength range into the fitting function to obtain the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function, comprises: performing a logarithmic transformation on the fitting function to obtain a logarithmically transformed fitting function; using the historical dicing process parameters and the roughness of the bare wafer after the historical wire dicing process in the target wavelength range, calculating the first empirical coefficient, the second empirical coefficient, and the third empirical coefficient in the fitting function from the logarithmically transformed fitting function using the least squares method. The method of claim 1, wherein detecting the roughness of a bare wafer after wire dicing in a target wavelength range comprises: For each sampling point on the surface of the bare wafer, acquiring raw measurement data regarding the wafer surface height at each sampling point using a single-point measurement scheme; Based on the raw measurement data of all sampling points on the surface of the bare wafer, acquiring the wave signal of each sampling point in the target wavelength range using a filter corresponding to the target wavelength range; Determining the roughness of the bare wafer in the target wavelength range based on the amplitude statistics of the wave signals of all sampling points in the target wavelength range. The method described in claim 5, wherein the filter is a circular filter and is composed of a double Gaussian filtering function; The radius of the filter's effective range is determined by the distance between the sampling point and the center of the bare wafer, and when the distance between the sampling point and the center of the bare wafer is less than or equal to a critical distance, the radius of the filter's effective range is a first radius; when the distance between the sampling point and the center of the bare wafer is greater than the critical distance, the radius of the filter's effective range is a second radius; wherein both the first and second radii are determined by the upper limit of the target wavelength range, and the first radius is greater than the second radius. The method of claim 5, wherein determining the roughness of the bare wafer in the target wavelength range based on the statistical values of the amplitudes of the wave signals at all sampling points in the target wavelength range comprises: dividing the bare wafer into multiple analysis regions; calculating the range of the amplitude of the wave signal in the target wavelength range for each analysis region based on the amplitude values of the wave signals at the sampling points in each analysis region; determining the average range of the amplitudes of the wave signals in the target wavelength range for all analysis regions as the roughness index of the bare wafer in the target wavelength range. The method as described in any of claims 1 to 7, wherein the roughness threshold is 1 nm when the target wavelength range is 0 to 1.8 micrometers; the roughness threshold is 3 nm when the target wavelength range is 1.8 to 22 micrometers; the roughness threshold is 25 nm when the target wavelength range is 22 micrometers to 20 millimeters. A wafer roughness improvement apparatus includes: a detection section, a comparison section, and a process improvement section; wherein, the detection section is configured to detect the roughness of a bare wafer after a wire dicing process in a target wavelength range; the comparison section is configured to compare the roughness with the set roughness threshold, and trigger the process improvement section when the roughness is greater than or equal to the set roughness threshold; the process improvement section is configured to select dicing process parameters to be adjusted based on a priority order of dicing process parameters indicated by a fitting function between multiple dicing process parameters and the roughness in the target wavelength range. A computing device comprising: a processor and a memory; the processor being configured to execute instructions stored in the memory to implement a method for improving wafer roughness as described in any one of claims 1 to 8. A computer-readable storage medium storing at least one instruction for execution by a processor to implement a method for improving wafer roughness as described in any one of claims 1 to 8. A wafer obtained by a wire dicing process improved by the wafer roughness improvement method as described in any one of claims 1 to 9, and having a roughness in a target wavelength range of at least one of the following: When the target wavelength range is 0 to 1.8 micrometers, the roughness is 0.01 to 1 nm; When the target wavelength range is 1.8 to 22 micrometers, the roughness is 1 to 3 nm; When the target wavelength range is 22 micrometers to 20 millimeters, the roughness is 3 to 30 nm.